This infact seems to be correct. After investigation it turns out that it is
lilo that baulks at anything bigger than 16MB of initrd (after
decompression). I have confirmed this using 2.2.20, 2.2.21, 2.4.7 and a few
printks to see if these images are actually getting the start_initrd value
passed. Anything bigger than 16MB and lilo will either:
a. Seem to complete the load but not pass the ramdisk_start value to the
kernel
or
b. Cold hang during the copy of the compressed image into memory
I've had a bit of a look at the lilo source but the assembly stuff is
meaningless to me. Can anyone please tell me if this is the intended
function of lilo (to be limited to a 16MB initrd?) or point me in the right
direction as to where I might start looking for this bug in the lilo source.
>From what Mike says, below it would seem that this problem is not specific
to Lilo, but effects Loadlin aswell. I will try grub next.
Thanks,
Andrew.
----- Original Message -----
From: "Mike Galbraith" <EFAULT@gmx.de>
To: "Andrew" <temp01@bluereef.com.au>
Sent: Friday, August 09, 2002 10:45 PM
Subject: Re: 2.2.20 ramdisk(initrd) not found by kernel
> Andrew wrote:
>
> >I have a problem that I have been wrestling with now for a number of days
> >with no solution, and I'm hoping someone can help.
> >
> >My problem is that when the 2.2.20 kernel loads, lilo doesn't seem to
load
> >the rootfs.img into RAM or it gets dropped before the kernel finds it.
That
> >is I don't get the
> >kernel message 'RAMDISK found at block 0' message and thus Linux panics
with
> >something like "root file system not found on dev 1:0".
> >
> >I have a stock 2.2.20 kernel with ramdisk and initrd support compiled in.
> >RAMdisk size is 64MB although I've also tried 32MB and 128MB.
> >I have tried kernel builds with module support and without (everything
> >compiled in)
> >I'm using the latest lilo I can find with the following config:
> >
> >boot=/dev/hdc
> >disk=/dev/hdc
> > bios=0x80
> >map=/map
> >install=/boot.b
> >backup=/boot.1600
> >prompt
> >linear
> >timeout=50
> >password=maintenance
> >restricted
> >image=/vmlinuz-2.2.20up
> > label=test
> > ramdisk=65536
> > initrd=/rootfs.img
> > root=/dev/ram
> >
> >The server is a uni processor PIII server with 512MB of RAM
> >
> >The sizes of my rootfs.img and kernel are:
> > 8713856 Aug 7 12:55 rootfs.img (this is an ext2 compressed image)
> > 787022 Aug 7 12:17 vmlinuz-2.2.20up (this is a monolithic bzImage
kernel)
> >
> >Lilo when building doesn't report any errors in fact it says it
successfully
> >maps the RAMdisk ok
> >
> >It almost seems like there is some finite size or block limit that my
> >rootfs.img+kernel is greater than, that stops the
> >RAMdisk being loaded or being found if it is infact being loaded at all.
> >
> >Is there a finite limit on the size initrd can be? Enlarging the ramdisk
> >size or altering the block size of the image doesn't seem to make any
> >difference.
> >
>
> 16MB was the largest I could ever load. (but that could have been a
> loadlin limitation)
>
> -Mike
>
>
>
>
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Thu Aug 15 2002 - 22:00:21 EST